Optimization Based Collision Avoidance for Multi-Agent DynamicalSystems in Goal Reaching Task

TL;DR

This paper introduces a distributed Model Predictive Control approach using ADMM for collision avoidance in multi-agent systems during goal-reaching tasks, emphasizing computational efficiency and solution reliability.

Contribution

It extends optimization-based collision avoidance to multi-agent systems with a distributed formulation and ADMM solution method.

Findings

Demonstrates effective collision avoidance in multi-agent transitions

Analyzes computational times and solution reliability

Provides a framework for distributed multi-agent control

Abstract

This work presents a distributed MPC-based approach to solving the problem of multi-agent point-to-point transition with optimization-based collision avoidance. The problem is formulated, motivated by the work on collision avoidance for multi-agent systems and dynamic obstacles. With modifications to the formulation, the problem is converted into a distributed problem with a separable objective and coupled constraints. The problem is divided into local sub-problems and solved using Alternating Directions Method of Multipliers(ADMM) applied on an augmented local lagrangian objective.This work aims to understand the multi-agent point-to-point transition problem as an extension of optimization-based collision avoidance and analyze the aspects of computational times, reliability, and optimality of the solution obtained.